Cement and Concrete Research 86 (2016) 78–84

Contents lists available at ScienceDirect

Cement and Concrete Research

j ourna l homepage: www.e lsev ie r .com/ locate /cemconres

Quantification of chloride in concrete samples using LA-ICP-MS

Maximilian Bonta a, Armin Eitzenberger a, Stefan Burtscher b, Andreas Limbeck a,⁎a TU Wien, Institute of Chemical Technologies and Analytics, Getreidemarkt 9/164-IAC, 1060 Vienna, Austriab TU Wien, Technische Versuchs- und Forschungsanstalt GmbH, Gutheil-Schoder-Gasse 17, 1230 Vienna, Austria

⁎ Corresponding author.E-mail address: andreas.limbeck@tuwien.ac.at (A. Lim

http://dx.doi.org/10.1016/j.cemconres.2016.05.0020008-8846/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 June 2015Accepted 9 May 2016Available online 20 May 2016

Inspection of the chloride content in concrete samples is an important factor for determination and planning ofmaintenance measures. The standard method given by European legislative for chloride determination is thetitrimetric Volhard method. However, it requires significant experimental effort, and knowledge about the ce-ment content in the concrete is required. In this work, laser ablation inductively coupled plasmamass spectrom-etry (LA-ICP-MS) is proposed as an alternative method to the commonly used wet-chemical approach. LA-ICP-MS allows analysis of drilled core concrete samples with increased depth resolution, minimal workload forsample preparation, and fast analysis time. Furthermore, exploiting the multi-elemental capabilities of ICP-MS,an in-depth analysis of the sample composition, allowing a differentiation between cement phase and aggregatesis feasible. Using this information, selective determination of chloride in the cement phase becomes possible,which helps to increase the reliability of the determined chloride concentrations.

© 2016 Elsevier Ltd. All rights reserved.

Keywords:Chloride (D)Concrete (E)LA-ICP-MSQuantificationCement (D)

1. Introduction

Free chloride can locally weaken the passivation layer of the rein-forcing steel [1], and the induced corrosion of the steel can severelyaffect structural stability. The chloride content near the reinforcementis therefore an essential parameter to determine and planmaintenancemeasures for concrete structures [2].

Concrete consists of multiple main components: aggregates (stones,rocks, or sand, in variablemixtures), water, admixtures, and cement as abinding agent. Even though the use of chloride containing aggregates ispermitted by European standards (EN 12620:2013), completelychloride-free environments are not economically realizable. Additional-ly, external sources of chloride, such as de-icing salts, play a consider-able role for chloride intake into the concrete structure.

The standard method for chloride determination specified by theEuropean standard EN 14629:2007 [3] is the titrimetric Volhardmethod, which exploits precipitation of solubilized chloride with silvernitrate. Concrete samples are collected by drilling holes into the struc-tures, and collecting either complete cores, or borehole dust. Cores aretypically cut into slices of 10–20 mm thickness, and ground for furtheranalysis. Borehole dust can be used directly without homogenization.The powdered samples are solubilized using nitric acid, and the chlorideconcentration in the solution is determined by titration. The cementcontent (the chloride concentration is referenced to the amount ofcement in the concrete) is either known from the concrete production,or estimated by conservative rules dictated by the standards [3]. Usingbothmethods for concrete sampling, depth profiles at depth resolutions

beck).

of around 10–20 mm can be collected. Higher depth resolutions can beobtained by specialized techniques (e.g., by grinding of slices). Howev-er, as the chloride content is determined as a chloride concentration inthe cement fraction, in such cases the inhomogeneity of the aggregateshas to be taken into account. The determination thereof will becomeless accurate with smaller sampling volumes (i.e., higher depth resolu-tion). Alternative methods for chloride determination in concretesamples are other titrimetric methods, such as potentiometric titration,or the use of ion selective electrodes. Also, more advanced instrumentalanalytical methods, such as X-ray fluorescence (XRF) [4], laser inducedbreakdown spectroscopy (LIBS) [5–8], or laser ablation inductivelycoupled plasma mass spectrometry (LA-ICP-MS) [9] have already beenused for the determination of chloride in concrete. Those methodsallow some improvements in depth resolution compared to titrimetricapproaches. In addition to improving the limits of detection and thespecificity of the analysis, the workload for such analyses could bereduced.

LA-ICP-MS is a method with a broad application range in theenvironmental and geological sciences [10]. Allowing for direct solidsample analysis, excellent detection limits for most elements, and awide linear range, this method is very powerful for elemental analysis.In this work, the analysis of concrete core samples using LA-ICP-MS ispresented. The objective of the presented work is to provide a moresensitive and more accurate determination of the chloride content inconcrete compared to the standard method. In a novel approach,selective quantification of chloride in the cement phase is performed.As only the chloride content in the cement phase is determined, thetotal chloride content can be correlated with the free chloride content,which is responsible for rebar corrosion. In preliminary experiments,in-house prepared concrete samples with different types of aggregates

79M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

and known chloride and aggregate contents were analyzed. Chloridecontaining aggregates are excluded from data evaluation and thus, avery reliable determination of the chloride content in the cementphase combined with high depth resolution can be obtained. Thedeveloped method was applied to drilled core samples from a retainingwall beside a city highway obtained from different sampling positions;the obtained depth profiles and the chloride quantification werecompared with results from conventional titration.

2. Material and methods

2.1. Sample preparation and collection of drilled core samples

In order to produce concrete sampleswith known aggregate contentand chloride concentration, some specimens were prepared in-house.Commercially available cement powder (baumit, Austria) was mixedin a ratio of 7:2 (v/v) with NaCl-containing water (conc. 3.79 g L−1),yielding a total chloride content of 0.5 m% in the cement mixture.Preliminary titrations showed that the used cement does not have aconsiderable chloride content (i.e., not quantifiable by titration).Approx. 60.0 m% aggregates were added to each concrete mixture, theexact aggregate content was determined by weighing. The used aggre-gates originated from different areas in Austria, resembling varyingcompositions: River gravel, limestone, and granite had a grain sizedistribution between 1.0 and 16 mm. Aggregates and cement werethoroughly mixed and molded into cylindrically shaped PE molds withan inner diameter of 50 mm. The concrete was allowed to harden for3 days at room temperature and 40% relative humidity until furtherprocessing.

Drilled core samples were taken from a retaining wall beside a cityhighway in Vienna (Austria). Sampling from the concrete structurewas performed at four different heights (0.8, 1.8, 2.6, and 4.6 m abovestreet level), facing the roadway; six core samples were taken at eachheight with a horizontal distance of 1.5 m between the samplingspots. All cores had a diameter of 45mm,with a length of approximately100 mm.

All cylinders, molded concrete samples as well as drilled coresamples were cut along the length of the sample into two semicircularhalves by dry cutting, in order not to mobilize chloride. Until furtherprocessing, the samples were stored at 25 °C and a relative humidityof 40% separately in plastic bags. Storage time was never longer thantwo weeks. The flat side of one half-cylinder was used for LA-ICP-MSanalysis. Parts of the other half of the molded concrete samples wereground using a tungsten carbide mortar to a maximum particle sizebelow 1 mm for subsequent titrimetric analysis.

2.2. Instrumentation

All measurements were performed using a quadrupole ICP-MSdevice (iCAP Q, ThermoFisher Scientific, Bremen, Germany) coupled toan NWR213 laser ablation unit (ESI, Fremont, CA). The connectionbetween the instruments was established using PTFE tubing. Thesample material was ablated under a constant stream of helium,

Table 1Typical instrumental parameters used for sample measurement.

Laser ablation ICP-

Wavelength 213 nm PlasPulse duration 4 ns CooLaser repetition rate 10 Hz AuxLaser beam diameter 250 μm ConLaser energy 1.38 mJ ScanLaser scan speed 40 μm s−1 DwLaser beam geometry Circular MonHe gas flow 0.4 L min−1 MasAr make-up flow 0.8 L min−1

which was mixed with argon as make-up gas before introduction intothe ICP-MS. Tuning of the instrumental performance was performeddaily using a glass standard (NIST612, trace elements in glass, NationalInstitute of Standards and Technology, Gaithersburg, MD) for highestintensity of the 115In signal. Additionally, the oxide ratio was monitoredby the 140Ce16O/140Ce ratio, which was below 1.9% for all experiments.Laser parameters and gas flows were optimized in preliminary experi-ments to yield the best achievable signal stability aswell as signal inten-sity for the sample measurements. To prevent detector saturation, lessabundant isotopes were chosen for the bulk elements expected in theconcrete samples. Polyatomic interferences were not expected, as onlymajor sample constituents were analyzed. Typical instrumental param-eters used for the experiments are listed in Table 1.

2.3. Chloride determination using titration

Chloride determination using the Volhard method was performedaccording to the European norm EN 14629:2007 [3]. 50 mL high puritywater dispensed from a Barnstead EASYPURE II water system(ThermoFisher Scientific, Marietta, OH) were added to 1–3 g of theground sample material, which was deposited into PTFE beakers andweighed exactly. After addition of 10 mL conc. nitric acid (p.a., Merck,Darmstadt, Germany), another 50mLof waterwere added. Themixturewas heated until boiling and kept boiling for at least 3 min. Aftercomplete cooling to room temperature, 5 mL silver nitrate solution(0.05 mol L−1, p.a., Merck, Darmstadt, Germany) was added. Thereaction mixture was stirred to ensure quantitative precipitation ofthe formed silver-chloride. Subsequently, freshly prepared ammoniumiron-(III)-sulfate solution was added and the excess of silver ions wastitrated using ammonium thiocyanate solution (0.05 mol L−1, Merck,Darmstadt, Germany) until color change from white to a brownish-red color. A blank sample was treated in the same way, but withoutaddition of ground concrete.

2.4. Preparation of cement standards for LA-ICP-MS analysis

Pressed pellets from cement mixed with powdered sodium chloridewere manufactured to be used as standards. Milling of sodium chlorideand homogenization of standard materials was performed using anultrasonic mill (MM 400, RETSCH, Haan, Germany). Titrimetric chlorideanalysis was performed to determine the chloride concentration of thecement. 3.0 g cement powder was treated in the same way as theconcrete samples described above. Color change of the mixtureappeared after addition of the same titrant volume as for the blank sam-ple, indicating that the used cement has a chloride concentration ofbelow 0.01 mg g−1. Additionally, a blank pressing was prepared toinvestigate the cement powder by LA-ICP-MS. The 35Cl signal was notdifferent from the gas blank, i.e., also by LA-ICP-MS, chloride was notdetectable in the blank cement.

For preparation of the standards, 2.0 g cement powder wasmixed with various amounts of previously milled sodium chloride(p.a., Merck, Darmstadt, Germany). After homogenization, 300 mg ofthe mixture were pressed using a pneumatic press at 10 bar for 30 s,

MS

ma power 1550 Wl gas flow 14.0 L min−1

iliary flow 0.8 L min−1

es Nining mode Peak hopping

ell time per isotope 10 msitored isotopes 13C, 25Mg, 27Al, 29Si, 35Cl, 42Ca, 49Ti, 57Fes resolution 300 m/Δm

80 M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

yielding round cement pellets with a diameter of 12 mm. Thisprocedure yielded standard pellets containing 0.077, 0.23, 0.49, and1.01 m% chloride and a blank pellet. The concentrations were chosento be in the expected concentration range of the concrete samples.

2.5. LA-ICP-MS measurement of standards and samples

Before analysis, loose dust was removed from the sample surfaceusing clean and dry pressurized air. After placing samples and standardsin the ablation chamber, it was purged for 20 min with helium beforestarting the analysis. Time resolved intensity data for the selectedisotopes were collected using Qtegra software, provided by the manu-facturer of the ICP-MS instrument. For analysis of standards, straightlines 5.0 mm in length were ablated; analysis of the samples wasperformed by ablating lines 37.0 mm in length. On each core sample,24 lines with a distance of 3.3 mm between lines were measured; thescan direction was perpendicular to the drilling axis for drilled coresamples. After every two samples, one standard was measured, tocontrol possible instrumental drifts. Analysis time for each core samplewas approx. 20 min.

For signal evaluation, the transient signal intensity profiles obtainedfrom linescans of the standards were divided into 18 segments withequal length. Data points in each segment were averaged and a blankvalue for the signal obtained without material ablation was subtracted.This procedure was found necessary to determine representative depthprofiles, as chloride concentrations showed to vary significantly on dis-crete sampling positions in the real samples.

As the ablation behavior between concrete samples and pressedcement standards might differ significantly, an internal standard wasemployed. This internal standard can compensate changes in materialablation due to variations in laser focus resulting from a rough surfaceof the concrete samples. Since calcium is a major constituent of cementwith rather constant concentrations in different cement types (around55–65 m% CaO), this element was considered being a suitable internalstandard. Minor variations in the Ca-concentration of different concretetypes were neglected due to the overall positive effect of the internalstandardization on signal stability. Thus, the measured chloride intensi-ties (35Cl) were normalized to the corresponding calcium signal (42Ca)as an internal standard. Subsequently, the values for the 18 segmentswere averaged to obtain representativemean values for each investigat-ed line.

2.6. Imaging experiments

Investigation of elemental distributions was performed usingconsecutive linescans over the sample surface. A laser beam diameterof 100 μm at 10 Hz repetition rate with 300 μm s−1 scan speed wasused. Image construction from the acquired time resolved intensityprofiles and further processing of the obtained elemental distributionimages was accomplished using ImageLab [11] (v.1.02, Epina GmbH,Pressbaum, Austria). Detailed information about image acquisition andconstruction can be found elsewhere [12]. Corresponding light micro-scopic images were obtained using a Leica DM2500M microscope(Leica Microsystems, Wetzlar, Germany) in reflective-light mode.

3. Results

3.1. Sample analysis using titration

3.1.1. Analysis of cement-sodium chloride mixtures used as standards forLA-ICP-MS analysis

For verification of the chloride concentrations in the standards, allcement-sodium chloride mixtures were analyzed using titration.Approximately 500 mg of each standard were exactly weighed andused for the analysis. The chloride determination was performed infour replicates for all standards. Between 5.0 and 20 mL of ammonium

thiocyanate solution were required until the endpoint of the titrationwas reached. The analysis yielded chloride contents with relativestandard deviations of below 5%. Considering a single standard devia-tion, none of the actual results differed from the desired concentrations.

3.1.2. Chloride content in the in-house prepared concrete samplesIn contrast to the standards (pure cement), cured concrete samples

do not only consist of cement, but also of aggregates. In case of the in-house prepared concrete samples, the chloride content, as well as theamount of aggregates in the concrete was exactly known. The usedcement did not contain detectable amounts of chloride. Thus, the chlo-ride content in the resulting cement could be adjusted by a given NaClconcentration in the water used for preparation of the concrete. Basedon this knowledge, a wet-chemical analysis of these samples wasperformed and the results were compared with the actual values.Yielded chloride concentrationswere 0.49± 0.03m% (granite as aggre-gate), 0.737 ± 0.02 m% (limestone as aggregate), and 1.03 ± 0.03 m%(river gravel as aggregate). Except from the granite sample, the chloridecontents differed significantly from the value of 0.50m% chloride in thecement fraction of the prepared concretes. To explain these deviations,all aggregate types were analyzed for their chloride content usingtitration. For granite, no significant acid-soluble chloride content(b0.01 mg g−1) could be found. Chloride concentrations for limestonewere 0.867 ± 0.08 m% and for river gravel 1.37 ± 0.05 m%. Whenconsidering the known aggregate amount of 60%, chloride contents inthe cement phase of 0.542 ± 0.05 m% (limestone) and 0.521 ±0.03 m% (river gravel) can be calculated. Taking slight changes inaggregate-cement ratio into consideration, these values correspondwell with the nominal chloride concentration of 0.50 m%.

3.2. LA-ICP-MS measurements

3.2.1. Calibrations for signal quantificationFor calibration of the signal intensities, pressed pellets from the

cement-sodium chloride mixtures were used. The standards containing0.077m% chloride were not used for the calibration but considered as asample for validation of the calibration function. Thus, the cementpellets containing 0.23, 0.49, and 1.01 m% chloride, as well as a blanksample were used to determine a calibration function for the LA-ICP-MS procedure. Intensity values for 35Cl between 3500 cps (blank) and180,000 cps (highest standard) were obtained. After signal normaliza-tion using 42Ca as an internal standard, a linear regression model wascalculated from the known chloride concentrations in the pressedpellets. The calibration yielded a correlation coefficient of 0.997. Thelimit of detection (LOD) and limit of quantification (LOQ) were deter-mined by adding three times (LOD) and nine times (LOQ) the standarddeviation of the blank to the signal of the blank. An LOD of 0.008m%andan LOQ of 0.021 m% chloride were obtained.

The standard with 0.077% Cl was considered as a sample withunknown chloride content for validation of the calibration function.Quantification resulted in a chloride content of 0.072 ± 0.006 m%(α = 0.95, n = 4). The found chloride content does not differ signifi-cantly from the actual concentration; thus, the validity of the quantifica-tion procedure was demonstrated.

3.2.2. Elemental distribution in the concrete samplesFor differentiation between aggregates and cement phase by

elemental patterns, the samples needed to be investigated in closer de-tail. Especially differences between the compositions of the aggregateswere a point of interest. Criteria for the differentiation between aggre-gates and cement phase were defined for automated distinction ofcement and aggregates using the time-resolved ICP-MS data. To enablereliable conclusions, elemental imaging of multiple sample areas on allthree in-house prepared concrete types was performed. Investigationof the elemental distributions on the concrete surface revealed deeperinformation about the components of the concrete, especially the

81M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

composition of the aggregates. All of the five shown elements could bedetected in every part of the three samples. However, their intensitiesvaried significantly based on the composition of the sample area ofinterest.

For granite (Fig. 1), silicon and aluminum can be identified as majorcomponents of the aggregates. This fact correlates well with thechemical composition of granite. Carbon and calcium are only detectedat background level on areas with aggregates. The chloride signal in thecement phase is around40,000 cps (approx. 0.5m%Cl) and the distribu-tion is very homogeneous. In contrast, in the aggregate phase onlychloride signals around 10,000 cps can be detected, indicating thepresence of chloride in a lower concentration in this aggregate type.

When limestone is used as aggregate (Fig. 2), calciumand carbon arethemain elements detected in the aggregate phase. Calcium can also bedetected in cement, however with significantly lower intensities(approx. three times lower) than in the aggregates. Silicon and alumi-num are solely detected in the cement phase, indicating the absenceof siliceous aggregates. The chloride distribution in the cement phaseis very homogeneous, with average signals of 40,000 cps (approx.0.5 m% Cl). However, chloride can also be detected in the aggregateswith intensities up to 40,000 cps (approx. 0.9 m% Cl according to thedetermination by titration). Discrepancies in concentration and signalintensity comparing cement phase and aggregates can be explained byvarying ablation behavior, material transport, and analyte ionization.This phenomenon is commonly referred to as matrix effect, leading tothe fact that cement powder used for chloride quantification in thecement phase is not a suitable calibration standard for chloride in theaggregates.

In the case of river gravel being used as aggregate (Fig. 3), two typesof aggregates can be found: On the one hand carbonates with highsignals of carbon and calcium, and on the other hand siliceous minerals

Fig. 1. Elemental distributions on an in-house prepared concrete sample using granite as aggreg13C (b), 27Al (c), 29Si (d), 35Cl (e), and 42Ca (f).

with high silicon and aluminum signals. Carbon and calcium cannot bedetected in the siliceous aggregates. Chloride is distributed evenly inthe cement phase with average signals of around 40,000 cps (approx.0.5 m% Cl). However, as in the case of limestone, it can also be detectedabove background level inmost aggregates. Signal intensities for 35Cl upto 50,000 cps have been found, indicating high abundance of chloride inthose aggregates. As for the limestone aggregate type, also in the case ofriver gravel, the absolute signal intensities dependon the ablatedmatrixmaterial. Therefore, chloride concentrations found in the aggregates bytitration do no correlate with the detected signal intensities whencompared to the calibration function using cement powder.

Notable concentrations of chloridewere found in all aggregate types.Thus, the results from the elemental distribution analysis underline theneed for a selective quantification method of chloride in the cementphase.

3.2.3. Selective quantification of chloride in the cement phaseBased on the findings from the elemental distribution analysis, it is

possible to use LA-ICP-MS for the differentiation between cement andaggregates. For the three investigated concrete/aggregate types, adistinction criterion using the signal intensities of aluminum and calci-um was found suitable. However, for other aggregate types it might benecessary to adjust the criterion accordingly. Areas with either lowaluminum or calcium signal (or both) would be denoted as aggregates.The complete range of data points acquired during 1 day of analysis wastaken into consideration for definition of the criterion. Themedian valueof all data points acquired for 27Al and 42Cawas calculated. These valueswere multiplied by empirically selected factors to define the thresholdintensities. Factors of 0.5 (27Al) and 0.3 (42Ca)were found to be suitable.For selective quantification of chloride in the cement phase, every datapoint was checked for their 27Al and 42Ca intensity. If either of the

atesmeasured using LA-ICP-MS, visual image (a), as well as the elemental distributions of

Fig. 2. Elemental distributions on an in-house prepared concrete sample using limestone as aggregates measured using LA-ICP-MS, visual image (a), as well as the elemental distributionsof 13C (b), 27Al (c), 29Si (d), 35Cl (e), and 42Ca (f).

82 M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

intensities was below the threshold, this data point would not be takeninto consideration for the chloride quantification. In Fig. 4, an elementaldistribution image of the river gravel sample after application of thethreshold is shown. Compared with the visual image, a good distinctionbetween aggregates (and pores) and the cement phase is provided.

For analysis of the in-house prepared concrete, 10 parallel lines witha distance of 5 mm were measured. An average was calculated, aschloridewas expected to be homogeneous in those samples. 35Cl signalswere between 40.000 and 50.000 cps for all samples, resulting in chlo-ride concentrations of around 0.5m% in the cement phase. Those valuesare in good agreementwith the expected chloride concentrations in theconcretes. For the drilled core samples, depth profiles with 3.3 mmdepth resolution were obtained, resulting in 24 data points per drilledcore sample. 35Cl signal intensities ranged between 4000 cps and300,000 cps. Chloride concentrations between 2.0 and 0.03 m% werecalculated.

4. Discussion

4.1. Comparison of the results from titration and LA-ICP-MS measurement

Titrimetric analysis yielded varying results for three concrete typesdespite the fact that the same cement with a chloride content of0.5 m% was used for all mixtures. As indicated by the elemental map-ping experiments, some aggregates contain considerable amounts ofchloride. Based on the results, the titration does not seem feasible forthe exact determination of chloride in the cement phase. The foundcarbonate minerals will readily dissolve during acidic treatment of thesample as a sample preparation step for the titrimetric chloride deter-mination. As shown in the elemental images, most of the carbonate ag-gregates also exhibit chloride signal significantly higher than the

background signal. These findings underline the weakness of the titri-metric analysis: Aggregates may be soluble under acidic conditions,and theymay also contain significant amounts of chloride, thus, contrib-uting to the determined chloride concentration in the cement. Addition-ally, the determination of the aggregate content in unknown concretemixtures may be problematic and wrong estimations might provokefurther errors.

However, the influence of acid-soluble aggregates on the deter-mined chloride content strongly depends on the analyzed concretetype: While LA-ICP-MS analysis and titration yielded consistent resultsin the case of granite (acid insoluble) as aggregate material, the valuesdiffered significantly for the aggregate type river gravel and limestone.A comparison of the determined chloride concentrations is given inFig. 5. As expected, LA-ICP-MS analysis of all three cement types yieldedconsistent results of around 0.5m% chloride in the cement phase (for allconcretes the same cement was used), the results of the Volhardtitration method varied significantly from each other.

The results of the two analysis methods only match for the granitesample. As indicated by the elemental mapping experiment, thisaggregate type does not contain acid-soluble constituents. The aggre-gates form a solid residue in the mixture and are not dissolved by thenitric acid used during sample pretreatment for titration. Thus, onlythe total chloride content in the cement phase is determined. As a com-parison, the other two aggregate types contain acid-soluble minerals(e.g., limestone), which itself contain chloride. Thus, the effective chlo-ride content determined by the Volhard titration overestimates thechloride content in the cement. LA-ICP-MSmeasurement deliver correctresults as the above described distinction criterion can be used to selec-tively quantify chloride in the cement phase. This will lead to correctresults, disregarding the chloride concentration in the aggregates andif they are acid-soluble or not.

Fig. 3. Elemental distributions on an in-house prepared concrete sample using river gravel as aggregatesmeasured using LA-ICP-MS, visual image (a), aswell as the elemental distributionsof 13C (b), 27Al (c), 29Si (d), 35Cl (e), and 42Ca (f).

83M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

4.2. Comparison of Cl diffusion profiles

Within this study depth profiles were measured from samplescollected at different heights. Averages from six samples taken at thesame height above the roadway were calculated; one data point wasacquired every 3.3 mm. The profiles for all four sample heights arepresented in Fig. 6.

The overall maximum chloride concentration is 1.11 ± 0.07 m%, forthe samples taken 0.8 m above the roadway. Generally, the maximumchloride concentrations decrease with increasing height of sampling.

Fig. 4. Application of the distinction criterion between cement phase and aggregates:Concrete with river gravel; microscopic image (a) and distribution image afterapplication of the threshold (b); areas detected as aggregates are marked red, concreteis marked in green.

The maxima of the chloride depth profiles for every height consistentlyoccur at a depth of 15–20 mm, not in the outermost region of thedrilled core sample. Significant differences between heights can beobserved, considering the standard deviation of six replicate samplemeasurements.

The influence of de-icing salts used in winter is evident by the veryhigh chloride concentrations found in samples taken at lower heights.Chloride-rich water from the roadway splashes onto the concretestructure and penetrates the material by diffusion. The amount ofspray water in contact with the concrete surface is greater at lower

Fig. 5. Comparison between the determination of chloride content by Volhard titrationand LA-ICP-MS.

Fig. 6. Depth profiles of drilled core samples taken at four different heights above theroadway (0.8, 1.8, 2.6, and 4.6 m) determined by LA-ICP-MS. The marked areas aredisplayed with regard to the standard deviation of the average values of six independentsamples from the same height.

84 M. Bonta et al. / Cement and Concrete Research 86 (2016) 78–84

heights, due to the closer proximity to the roadway,where the chloride-rich water originates from. The samples investigated in this study wereremoved from the structure in the mid of November 2012. The lastfreezing day before sampling – and thus, probably the last use of de-icing salts – has been in early March 2012, eight months earlier. It canbe speculated that chloride from the outermost layers of the structurehas been washed away by rain and splash water from the roadway(with negligible amounts of chloride) during the months where node-icing salts have been used. This might explain the depth profileswith their maximum chloride content in deeper layers of the material.Consequently, the chloride content (especially in the outer concretelayers) and the found depth profiles might also be dependent on theseason and on the amount of de-icing salts used during the last winterperiods.

5. Conclusion

In the presented study, a fast and reliable analysis strategy fordetermination of the chloride content in concrete structures could bedeveloped. It was demonstrated that mineralization and subsequenttitration of the concrete samples can lead to systematic errors. The sys-tematic error results from acid-soluble chloride containing aggregates,other deviations may originate from an unknown aggregate content inthe samples. While titrimetric analysis yielded inconsistent results forthe chloride concentrations of three in-house prepared concrete

samples, LA-ICP-MS analysis resulted in a goodmatchwith the expectedchloride content.

The proposed LA-ICP-MS approach helps to increase the reliability ofthe obtained results by selectively addressing three major problems:

- Selective determination of chloride in the cement phase; no wrongresults due to dissolving aggregates that contain chloride; no errorsdue to incorrect determination of the cement-concrete ratio;

- High depth resolution of the chloride profiles that allows an exactassessment at the reinforcement depth and to use data for forecastmodels;

- The detection of localized maximum chloride concentrations whichare inaccessible to the titrimetric method is possible.

Acknowledgment

The authors acknowledge the financial support of ASFiNAG(Autobahnen- und Schnellstraßen-Finanzierungs Aktiengesellschaft;project P 30388, KOA640100) for the study of the chloride content in ce-ment by applying the new LA-ICP-MS method. Special thanks also toFritz Binder for support and fruitful discussions.

References

[1] U. Angst, B. Elsener, C.K. Larsen, Ø. Vennesland, Critical chloride content inreinforced concrete — a review, Cem. Concr. Res. 39 (2009) 1122–1138.

[2] L. Bertolini, B. Elsener, E. Redaelli, R. Polder, Corrosion of Steel in Concrete:Prevention, Diagnosis, Repair, second ed., 2013.

[3] EN 14629:2007, Products and systems for the protection and repair of concretestructures. Test methods, Determination of Chloride Content in Hardened Concrete,in, BSI 2007, p. 14.

[4] E. Proverbio, F. Carassiti, Evaluation of chloride content in concrete by x-rayfluorescence, Cem. Concr. Res. 27 (1997) 1213–1223.

[5] F. Weritz, S. Ryahi, D. Schaurich, A. Taffe, G. Wilsch, Quantitative determination ofsulfur content in concrete with laser-induced breakdown spectroscopy,Spectrochim. Acta B At. Spectrosc. 60 (2005) 1121–1131.

[6] T.A. Labutin, A.M. Popov, S.M. Zaytsev, N.B. Zorov, M.V. Belkov, V.V. Kiris, S.N. Raikov,Determination of chlorine, sulfur and carbon in reinforced concrete structures bydouble-pulse laser-induced breakdown spectroscopy, Spectrochim. Acta B At.Spectrosc. 99 (2014) 94–100.

[7] S. Eto, T. Matsuo, T. Matsumura, T. Fujii, M.Y. Tanaka, Quantitative estimation ofcarbonation and chloride penetration in reinforced concrete by laser-inducedbreakdown spectroscopy, Spectrochim. Acta B At. Spectrosc. 101 (2014) 245–253.

[8] G. Wilsch, F. Weritz, D. Schaurich, H. Wiggenhauser, Determination of chloridecontent in concrete structures with laser-induced breakdown spectroscopy, Constr.Build. Mater. 19 (2005) 724–730.

[9] N. Silva, T. Luping, S. Rauch, Application of LA-ICP-MS for meso-scale chlorideprofiling in concrete, Mater. Struct. 46 (2013) 1369–1381.

[10] N.S. Mokgalaka, J.L. Gardea-Torresdey, Laser ablation inductively coupled plasmamass spectrometry: principles and applications, Appl. Spectrosc. Rev. 41 (2006)131–150.

[11] http://www.imagelab.at.[12] M. Bonta, H. Lohninger, M. Marchetti-Deschmann, A. Limbeck, Application of gold

thin-films for internal standardization in LA-ICP-MS imaging experiments, Analyst139 (2014) 1521–1531.